PROJECT SUMMARY Learning is a complex process, and likely involves many areas of the brain that detect and process sensory inputs, integrate experience, and display behavior. Consistently, various neurological diseases that impair different brain areas are associated with profound defects in learning. Thus, bridging different spatial scales and understanding the dynamics of different brain regions are essential to understanding how learning occurs and potentially designing strategies to mitigate learning deficiency. However, it is currently not possible to achieve these goals in most experimental systems, and our understanding of learning is limited by the technical approaches by which either local circuit and cellular properties or coarse psychophysical parameters underlying learning are measured. Here, we propose to address these fundamental questions in a reduced system ? the nervous system of the nematode C. elegans. The rationale is that the wiring and genetic make-up of this network are well known, probing whole-brain dynamics with single-cell resolution with exquisite temporal resolution is technically ready for C. elegans, and the fundamental principles for the development and the function of the nervous system are well conserved between C. elegans and more complex animal models. Further, C. elegans exhibits many forms of learning, similar to those displayed by higher organisms in behavioral characteristics and molecular cellular underpinnings. Particularly, we will use an olfactory learning paradigm whereby C. elegans learns to avoid the odorants of pathogenic bacteria, a type of learning similar to the Garcia effect through which many animals, including humans, learn to avoid the smell and/or taste of a food that makes them ill. Our long-term goal is to understand how learning is encoded and executed by the function of the whole brain, and to inform the design of potential therapeutic strategies. The central hypothesis of this project is that learning engages global activity and the learned information is encoded in distinct functional modules. Specifically, we will test whether learned information is encoded in the learning-dependent changes in the activity patterns of individual functional modules and/or the interactions among the modules. To this end, we aim to image and analyze multi-cell and whole-brain dynamics under naive and learned conditions to characterize how learning alters the structure of the brain activities; further, we will introduce perturbations to the whole-brain dynamics and examine the consequences for learning. This work is innovative because (1) it brings a conceptual advance to understanding learning across scales, (2) it introduces technical advancement in whole-brain imaging and analyses, and (3) it demonstrates perturbation strategies for altering whole-brain dynamics that have behavioral consequences. It is significant, because it tests several highly plausible and likely conserved cellular and whole-brain dynamic models for learning and examine their behavioral consequences, it informs and facilitates learning studies in other systems, and it paves the way for designing interventions.